IL-23 Compensates for the Absence of IL-12p70 and Is Essential for the IL-17 Response during Tuberculosis but Is Dispensable for Protection and Antigen-Specific IFN-γ Responses if IL-12p70 Is Available¹

Tuberculosis (Tb)³ is the most prevalent infectious disease worldwide and kills >3 million people/year (1). Cell-mediated immune responses are critical to host defense, with IL-12 induction of IFN-γ production providing the primary pathway by which bacterial growth is controlled (2, 3). IL-12p70, is a heterodimeric cytokine made up of two disulfide-linked subunits, p35 and p40 (4), the biological activity of which requires its interaction with the IL-12R complex, composed of IL-12Rβ1 and IL-12Rβ2 (5).

In the absence of the IL-12p40 subunit, there is a profound loss of the protective immune response to mycobacteria (6, 7), which results in increased bacterial growth, reduced production of IFN-γ, decreased Ag-specific inflammation, and increased mortality (7). In contrast, mice lacking the IL-12p35 subunit are less susceptible to Mycobacterium tuberculosis (Mtb) infection. Specifically, they are able to inhibit bacterial growth to a limited degree, they can generate an Ag-specific IFN-γ T cell response and elaborate Ag-specific inflammation, and they can survive longer than the p40-deficient mice (7). The fact that both the p35- and p40-deficient mice lack IL-12p70, yet the p40-deficient mice are more susceptible to disease, implicates other p40-dependent molecules as mediators of protection in tuberculosis. A novel, four-α helix molecule, p19, has been identified that is structurally similar to p35 and forms a disulfide-bonded heterodimer with the p40 subunit (8); the p40/p19 heterodimer is termed IL-23.

Macrophages and dendritic cells (DC) produce IL-23 in a TLR2-dependent manner (9). Although IL-12p70 acts on naive T cells, IL-23 has been shown to induce proliferation and cytokine production in T cells with an activated phenotype (8, 10, 11). As an activator of STAT-4 and an inducer of IFN-γ production in human T cells, IL-23 was initially considered a pale imitator of IL-12p70 (8, 10). In murine cultures, however, IL-23 acts on T cells via STAT-3/STAT-4 heterodimers to promote the production not of IFN-γ, but of IL-17 (11, 12). This induction has been postulated to be a new axis of T cell differentiation, because neither IL-12p70 nor IL-4 is able to induce IL-17 production in T cells (13, 14, 15).

Both IL-23 and IL-17 are implicated in T cell-dependent inflammatory responses. Specifically, both IL-17- and IL-23-deficient mice have reduced Ag-specific inflammatory responses (14 , 16), whereas transgenic overexpression of IL-23 results in systemic inflammation and premature death of mice (17). Similarly, IL-17 has been implicated in T cell-dependent inflammatory responses by induction of cytokines and CXC chemokines by the interstitium (reviewed in Refs. 12 and 13). Most recently, the IL-23/IL-17 axis has been identified as the primary mediator of tissue damage in the murine experimental autoimmune encephalomyelitis model (15, 18) and in collagen-induced arthritis (19). Furthermore, the resistance of IL-23-deficient mice to collagen-induced arthritis is associated with abrogated T cell-dependent IL-17 production (19). Inflammation and granuloma formation are important components of mycobacterial disease progression; thus, IL-23 and IL-17 may be implicated in the immunopathologic response to Tb. In this regard, the relative deficiency of the Ag-specific inflammatory (7) and granulomatous (3) responses in B6.p40^−/− mice compared with that in B6.p35^−/− mice suggests that cellular accumulation during Tb may depend upon the IL-23/IL-17 pathway.

To determine the role of IL-23 in the immune response to Mtb, we aerogenically infected mice lacking the IL-23p19 subunit and followed disease progression by enumeration of bacteria, histological analysis, and assessment of Ag-specific CD4 T cell responses. We also determined whether and how IL-23 mediates the observed IL-12p70-independent, IL-12p40-dependent protection seen in IL-12p35-deficient mice by challenging mice lacking both IL-12p35 and IL-23p19.

C57BL/6 (B6), IL-12p35 gene-deficient (B6.p35^−/−), and IL-12p40 gene-deficient (B6.p40^−/−) mice were purchased from The Jackson Laboratory. Both male and female mice between the ages of 6 and 12 wk were used; experimental mice were age and sex matched. IL-23p19 gene-disrupted mice (B6.p19^−/−) were generated at Genentech by Drs. F. de Sauvage and Ghilardi as previously described (14). Specifically, the entire IL-23p19-coding region was replaced with an enhanced GFP. The B6.p19^−/−B6.p35^−/− double-gene-deficient mouse was generated by crossing B6.p35^−/− with B6.p19^−/−, typing the progeny by PCR using primers available from The Jackson Laboratory, and as previously reported (14); progeny from the F₃ generation were used. OT-II αβTCR-transgenic male mice, which are MHC class II I-A^b restricted and specific for OVA_323–339 (20), were obtained from the Trudeau Institute animal breeding facility.

An initial stock of H37Rv strain of Mtb was received from Dr. R. North (Trudeau Institute) and was grown in Proskauer Beck medium containing 0.05% Tween 80 to midlog phase and frozen in 1-ml aliquots at −70°C. For aerosol infections, subject animals were infected using a Glas-Col airborne infection system as previously described in detail (21).

Infected mice were killed by CO₂ asphyxiation, and the organs were aseptically excised. Each of the organs was individually homogenized in physiological saline, and serial dilutions of the organ homogenate plated on nutrient 7H11 agar. Bacterial colony formation was counted after 3 wk of incubation at 37°C (21).

Lung cell suspensions were prepared by perfusing cold saline containing heparin through the heart until the lungs appeared white, whereupon they were removed and sectioned in ice-cold DMEM (Mediatech-Cellgro) using sterile razor blades. Dissected lung tissue was then incubated in DMEM containing collagenase IX (0.7 mg/ml; Sigma-Aldrich) and DNase (30 μg/ml; Sigma-Aldrich) at 37°C for 30 min (21). Digested lung tissue was gently dispersed by passage through a 70-μm pore size nylon tissue strainer (Falcon; BD Biosciences); the resultant single-cell suspension was treated with Gey’s solution to remove any residual RBC, washed twice, and counted (21). Cells prepared in this way were used for ELISPOT and flow cytometric analyses.

Detection of Ag-specific IFN-γ- and IL-17-producing cells from infected lungs was conducted using an ELISPOT assay. In brief, cell culture plates (Multiscreen-HA; Millipore) were coated overnight at 4°C with monoclonal purified anti-mouse IFN-γ (clone R4-6A2; eBioscience) or monoclonal purified anti-mouse IL-17 (clone 50101.111; R&D Systems) in PBS. The plates were then incubated with blocking solution (DMEM containing 100 U/ml penicillin, 100 U/ml streptomycin, and 10% FBS; all additives from Sigma-Aldrich). Cells from the organs of infected mice were generated as described above and seeded at an initial concentration of 1 × 10⁵ cells/well in the Ab-coated plates; doubling dilutions of these cells were then made. Irradiated splenocytes from uninfected B6 mice were used as APCs at a concentration of 1 × 10⁶ cells/well in all wells. The relatively immunodominant (4% of CD4⁺ T cells in infected lungs) Mtb ESAT-6_1–20 peptide (10 μg/ml) (22, 23) was used as the Ag for assays from infected mice, whereas the OVA peptide was used to stimulate cells from OT-II αβTCR-transgenic mice; mouse rIL-2 (Sigma-Aldrich; 10 U/ml) was added to all wells. After 24 h of incubation in 5% CO₂ at 37°C, the plates were washed, and biotinylated anti-mouse IFN-γ (clone XMG1.2; eBiosciences) or biotinylated anti-mouse IL-17 Ab (clone AUS02; R&D Systems) was used to detect the captured cytokine. Spots were visualized using streptavidin-alkaline phosphatase (DakoCytomation) and 5-bromo-4-chloro-3-indoylphosphate/nitro blue tetrazolium (Sigma-Aldrich) as substrate. Spots were enumerated visually under a dissection microscope. The frequency of responding cells was determined and applied to the number of cells per sample to generate the total number of responding cells per organ. Neither cells cultured in the absence of Ag nor cells from uninfected mice produced detectable spots.

Single-cell suspensions from the lungs of infected mice were prepared as described above and cultured with anti-CD3 and anti-CD28 in the presence of monensin as previously described (7). After 4 h, cells were stained with labeled Abs specific for CD4 (clone GK1.5) and CD3 (clone 145-2C11). Cells were then fixed and permeabilized, and the presence of intracellular IFN-γ was determined using clone XMG 1.2. Isotype Abs were used in parallel samples to confirm the specificity of binding. Cells were analyzed using CellQuest on a FACSCalibur, dual-laser, flow cytometer (BD Biosciences) with excitation at 488 and 633 nm. Lymphocytes were gated based on their forward and side scatter characteristics and then on CD4 and CD3, and the number of such lymphocytes per organ was determined. CD4-positive lymphocytes were analyzed for their expression of IFN-γ, and the frequency was applied to the total number of CD4 T cells per organ to determine the number of IFN-γ-producing cells per organ.

Lung tissue from infected and control mice was homogenized and frozen in 1 ml of TRIzol reagent (Invitrogen Life Technologies), and total RNA was extracted according to the manufacturer’s protocol. RNA samples (n = 4) from each group and each time point were treated with DNase (Ambion) and reverse transcribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen Life Technologies). cDNA was then amplified using TaqMan reagents on the ABI PRISM 7700 sequence detection system (Applied Biosystems). Samples were run in the absence of the RT enzyme to confirm that signal was derived from RNA. The fold increase in signal over that derived from uninfected tissue or cells was determined using the ΔΔct calculation recommended by the ABI PRISM 7700 manufacturer. The primer and probe sequences for murine IL-12p40, IL-12p35 (24), IL-23p19 (7), GAPDH, IFN-γ, NO synthase 2 (nos2), and TNF-α (25) were previously published. The primer and probes for LRG-47 were obtained from Applied Biosystems. The IL-17 probe/primer set was designed at the Molecular Biology Core Facility, Trudeau Institute, and had the following sequences: forward primer, CCACGTCACCCTGGACTCTC (exon 1; 185–204 bp); reverse primer, CTCCGCATTGACACAGCG (exon 2; 268–285 bp); and probe, CCTCTGTGATCTGGGAAGCTCA-GTGCC (exon 2; 233–259 bp). All primer and probes sets were determined to be 95% efficient using murine mRNA by the Molecular Biology Core Facility, Trudeau Institute.

The caudal lobe of each lung was inflated with 10% neutral buffered saline and processed routinely for light microscopy. Sections were then stained with H&E and assessed by histological techniques as described previously (26). Four to eight mice were examined for each group and at each time point. Sections were screened and scored in a blinded manner by a qualified veterinary pathologist (K. Sakamoto).

DC were generated from the bone marrow cells of B6, B6.p19^−/−, B6.p40^−/− or B6.p35^−/− mice as previously described (27). Briefly, cells were extracted from mouse femurs, and 1 × 10⁶ BMDC were cultured in 10 ml of DMEM supplemented with 10% FBS (complete DMEM (cDMEM)) containing 20 ng/ml murine rGM-CSF (rmGM-CSF; PeproTech). Cells were cultured for 3 days at 37°C in 5% CO₂, after which an additional 10 ml of cDMEM containing 20 ng/ml rmGM-CSF was added. On days 6 and 8, half the volume of culture medium was removed, the culture was centrifuged, and the cell pellet was added back to the same plate in 10 ml of fresh cDMEM containing 20 ng/ml rmGM-CSF. On day 10 of culture, nonadherent cells were collected by centrifugation, counted, and used as APCs. BMDC were characterized morphologically by cytospins, and >80% of the cells contained dendrites.

BMDCs from 10-day cultures were placed in six-well plates, infected with Mtb H37Rv at a multiplicity of infection of 10 in antibiotic-deficient cDMEM, and incubated at 37°C in 5% CO₂. At different times after infection, plates were centrifuged, and culture supernatant was taken from each well and frozen for analysis by ELISA. The remaining adherent and nonadherent cells were treated with TRIzol, RNA was extracted, and the fold induction of cytokine mRNA levels was determined by real-time PCR.

ELISA Ab pairs were used to detect IL-12p40 and IL-12p70 (BD Pharmingen) in the culture supernatants. Recombinant cytokines were used to generate standard curves.

Naive CD4⁺ T cells were isolated from single-cell suspensions of spleens and lymph nodes of OT-II αβTCR-transgenic mice using magnetic CD4⁺ beads (clone GK1.5; Miltenyi Biotec) on an AutoMACS machine following the manufacturer’s instructions. We routinely obtained >95% CD4⁺ T cells. Mtb-infected BMDCs from B6, B6.p35^−/−, B6.p19^−/−, and B6.p40^−/− mice were generated as described above. Naive OT-II CD4⁺ T cells (3 × 10⁵ cells/ml) were cultured with infected DCs (3 × 10⁵ cells/ml), the antigenic peptide OVA_323–339 (5 μM), and IL-2 (10 ng/ml) for 72 h at 37°C in 5% CO₂. The primed cells (5 × 10⁵ cells/ml) were then washed, counted, and restimulated in the presence of irradiated B6 splenocytes (5 × 10⁵ cells/ml) and OVA_323–339 peptide (5 μM) for an additional 72 h at 37°C in 5% CO₂. Lymphocytes were again washed and counted, and the frequency of OVA_323–339-specific, IFN-γ- or IL-17-producing CD4⁺ T cells was determined by ELISPOT.

Differences between the means of experimental groups were analyzed using Student’s t test. Differences were considered significant at p ≤ 0.05.

The potential for IL-23 to mediate a protective immune response to Tb was suggested by the induction of IL-23p19 in Mtb-infected lungs (7), the differential susceptibility of B6.p35^−/− and B6.p40^−/− mice to Tb (7), and the published role of IL-23 in mediating inflammatory responses (14, 16, 17). To determine whether IL-23 is intrinsically protective, the bacterial burden in B6 and B6.p19^−/− mice was compared over time. Surprisingly, there were no significant differences in the bacterial burden of any of the organs screened, apart from day 140 in the spleen, where a significantly higher bacterial burden was found in the absence of IL-23p19 (Fig 1). The absence of substantial differences in the bacterial burden suggests that neither IL-23 nor any other p19-dependent molecule is required to mediate either initial cessation or long-term control of mycobacterial growth in the murine model.

The ability of the B6.p19^−/− mice to control bacterial burden was surprising, because IL-23 was initially identified for its role in proliferation and IFN-γ production by activated CD4⁺ T cells (8). To determine whether the absence of IL-23 affected the Th1 response to Tb, we compared this response in Mtb-infected B6 and B6.p19^−/− mice. In light of the bacterial numbers, we were not surprised to find comparable numbers of CD4⁺ T cells specific for the immunodominant mycobacterial Ag, ESAT6_1–20 (23), in the lungs of both groups of mice throughout 140 days of infection (Fig. 2,a). We also found the total number of CD4⁺ T cells capable of expressing IFN-γ in the lung after a polyclonal TCR stimulus (anti-CD3/anti-CD28) to be equivalent between the two groups (Fig. 2,b). Furthermore, in confirmation of a failure of IL-23 to affect the total IFN-γ response in the lung, we demonstrate in this study that equivalent levels of IFN-γ mRNA were detectable in the lungs of both groups throughout the infection (Fig. 2 c). This apparent inability of the absence of IL-23 to compromise the IFN-γ response in the lung suggests that IL-23 is dispensable for this component of the protective response to Tb.

As discussed above, IL-23 has been implicated in the induction of an IL-17-producing phenotype in CD4⁺ T cells (11, 12, 15). To determine first whether there is an IL-17 response to Tb and second whether IL-23 affects this response, we compared the number of Ag-specific, IL-17-producing CD4⁺ T cells produced in the presence and the absence of IL-23. We show in this study that IL-17-producing Ag-specific CD4⁺ T cells are induced during pulmonary infection with Mtb (Fig. 3,a), and that the number of these cells was significantly reduced in B6.p19^−/− mice compared with B6 mice throughout the infection (Fig. 3,a). In support of a global inhibition of IL-17 production within the IL-23-deficient infected lung, we also found that although B6 mice expressed IL-17 mRNA early upon infection and maintained this expression throughout infection, B6.p19^−/− mice failed to significantly express IL-17 mRNA throughout the infection (Fig. 3 b).

Having identified significant reductions in the levels of IL-17-producing, Ag-specific CD4⁺ T cells and total IL-17 mRNA in the Mtb-infected lung, we wanted to determine what effect this had on the induction of specific genes in the lung. Two known antimycobacterial agents, the inducible NOS gene (nos2) (28) and the LRG-47 gene (29) (a member of the 47-kDa GTPase family involved in maturation of phagosomes), were both induced in the lungs upon infection. Induction was comparable between B6 and B6.p19^−/− mice throughout infection (Fig. 4). These observations support those made above that the IFN-γ-dependent antimycobacterial immune response was not compromised in the absence of IL-23. As previously reported, the IL-23p19 subunit was induced in the lungs of infected mice (data not shown). Both the p35 and p40 subunits of IL-12 were also induced during infection, and the levels of induction were not affected by the absence of IL-23 or IL-17 (data not shown).

As discussed above, both IL-23 and IL-17 are potent inducers/mediators of inflammatory responses (30, 31, 32). In the absence of IL-23 and IL-17, the difference in the granulomatous response to Mtb was not dramatic. As shown in Table I, the absence of IL-23 and IL-17 resulted in the more rapid appearance of moderate inflammation, with severe inflammation as an end point. This inflammation corresponded with a minor increase in the consolidation of the IL-23-deficient lung response and with a slightly reduced deposition of fibrin. IL-17 has been implicated in neutrophil recruitment; however, there was no difference in the accumulation of granulocytes between the two groups of mice (data not shown).

Despite the apparent failure of IL-23 to contribute to the protective response seen in mice capable of making IL-12p70, we were still interested in whether this cytokine contributed to the IL-12p70-independent, IL-12p40-dependent protection seen in B6.p35^−/− mice (7). To determine whether IL-23 contributed to this protection, we generated mice lacking both IL-23p19 and IL-12p35 and infected them aerogenically with Mtb. We show in this study that in the absence of IL-23, mice lacking IL-12p35 are unable to express the moderate level of protection seen in the presence of IL-23. Specifically, there was a significant difference in the bacterial burden in the lung (Fig. 5,a), liver (Fig. 5,c), and draining lymph nodes (Fig. 5,d) between B6.p35^−/− mice and B6.p19^−/−p35^−/− mice; the difference in bacterial burden in the spleen approached significance (Fig. 5,b). As previously reported (7) B6.p40^−/− mice supported significantly more bacterial growth than did B6.p35^−/− mice in all organs (Fig. 5).

To examine how the absence of IL-23 affects the immune response of Mtb-infected B6.p35^−/− mice, we compared the induction of mRNA within the lungs of Mtb-infected, gene-deficient mice. We show in this study that although B6.p35^−/− mice exhibited similar levels of mRNA for the protective molecules NOS2 (Fig. 6,a) and LRG47 (Fig. 6,b) to the B6 mice (Fig. 6), the B6.p19^−/−p35^−/− mice exhibited similar mRNA levels as B6.p40^−/− mice; these levels were significantly lower than those in both B6 and B6.p35^−/− mice. We measured other cytokines, such as IL-10 and IL-18, and although these cytokines were induced upon infection, the expression level was not altered by the absence of IL-12p40, IL-12p70, or IL-23 (data not shown).

To determine whether IL-23 was acting to induce Th1 cells during Mtb infection in B6.p35^−/− mice, we used the ESAT-6_1–20 peptide-based ELISPOT to compare the numbers of IFN-γ-producing CD4⁺ T cells in the lungs of infected, gene-deficient mice. We show in this study that although B6.p35^−/− mice can generate a small, but measurable, Ag-specific Th1 response to Mtb, B6.p19^−/−p35^−/− mice generate a significantly reduced response, which is not different from that seen in B6.p40^−/− mice (Fig. 7,a). We also wanted to determine the effect of IL-12p70 on the development of the Mtb-specific, IL-17-producing CD4⁺ T cells. We show in this study, using the ESAT-6_1–20 peptide-driven ELISPOT, that the number of these cells is significantly increased in the lungs of Mtb-infected B6.p35^−/− mice compared with B6 mice (Fig. 7 b) and that this response is ablated in the B6.p19^−/−, B6.p40^−/−, and B6.p19^−/−p35^−/− mice.

We show that in the absence of IL-23, the frequency and total number of mycobacteria-specific, IL-17-producing CD4⁺ T cells are profoundly reduced (Figs. 3,a and 7b). Although the role of IL-23 in the expression of an IL-17-producing phenotype has been previously documented (11, 14, 18), in these publications IL-23 was relatively inefficient at driving naive CD4 T cells to an IL-17-producing phenotype. The in vivo data (Figs. 3,a and 7b) suggest that Mtb infection is a potent inducer of IL-17-producing CD4⁺ T cells, and we therefore determined whether Mtb was able to drive DC to a state that would efficiently induce IL-17-producing, Ag-specific CD4⁺ T cells. First, we determined that IL-12p40 (Fig. 8,a), IL-12p35 (Fig. 8,b), and IL-23p19 (Fig. 8,c) mRNA were all induced during infection of DC with live Mtb, and that this expression was maintained at high levels up to 48 h (Fig. 8). Both IL-12p40 and IL-12p70 protein were induced to detectable levels from 4 through 48 h (500–3500 pg/ml IL-12p40 and 100–300 pg/ml IL-12p70). We then used infected C57BL/6 and gene-deficient DC to prime naive CD4⁺ TCR-transgenic T cells (OT-II) in the presence of their cognate Ag (OVA_323–339). This priming was designed to determine the ability of Mtb-matured DC to induce naive CD4⁺ T cells to an IL-17-producing phenotype as well as the role of IL-23 in that induction. In this study we show that when CD4⁺ T cells are primed in the presence of Mtb-infected DC, a substantial number of cells express an IL-17-producing phenotype after restimulation with Ag in the presence of irradiated uninfected splenocytes (Fig. 8,d). If these cells are primed in the absence of IL-23, however (i.e., B6.p19^−/− DC), there is a significant drop in the number of IL-17-producing cells despite the fact that restimulation is with IL-23-competent splenocytes (Fig. 8,d); the number of IFN-γ-producing cells is unaffected (Fig. 8,e). IL-12p70 has been shown to limit the production of IL-17 in culture (11); thus, we determined how the absence of IL-12p70 affected the generation of IL-17-prodcuing CD4⁺ T cells from naive T cells. We show in this study that when naive CD4⁺ T cells are primed in the absence of IL-12p70 (i.e., by B6.p35^−/− DC), a significantly increased number become IL-17-producing cells compared with cells primed in the presence of IL-12p70 (Fig. 8,d). In the absence of IL-12p40, there was very limited induction of IL-17-producing CD4 T cells (Fig. 8 d).

We show in this study for the first time that the absence of the IL-23 p19 subunit has little effect on disease progression during both early and chronic Mtb infection. We also show that in the absence of IL-23, Ag-specific IFN-γ responses are not compromised and are able to adequately induce antimycobacterial activity. The absence of IL-23 does result in a profound reduction in the frequency and number of Ag-specific, IL-17-producing CD4⁺ T cells and in local IL-17 mRNA production in the lung; however, this fails to substantially alter the immunopathologic response to Mtb in the lung. This limited phenotype suggests that although the IL-23/IL-17 axis of T cell activation is induced in tuberculosis, it does not mediate antibacterial or profound inflammatory elements of the primary host response. In contrast, however, we demonstrate that when IL-12p70 is absent, as is the case in mice lacking the IL-12p35 gene, then IL-23 is required for the generation of Ag-specific, IFN-γ-producing CD4⁺ T cells, induction of anti-mycobacterial genes, and control of bacterial growth. The increased numbers of Ag-specific, IL-17-producing CD4⁺ T cells seen both in vitro and in vivo suggest that IL-12p70 regulates the activity of IL-23 during murine tuberculosis.

The fact that IL-23 deficiency does not adversely affect the Ag-specific IFN-γ response during tuberculosis is not surprising based on the growing literature showing that the primary function of IL-23 is the induction of Ag-specific CD4⁺ T cells that produce IL-17, IL-6, and TNF (11, 14, 15, 19). IL-23 was initially identified as being able to induce IFN-γ production in already activated human CD4⁺ T cells (8); however, the ability of IL-23 to induce IFN-γ-producing cells in the mouse is less clear. Our previous data demonstrating that IL-12p35-deficient mice were able to generate an Ag-specific IFN-γ response to Mtb (7) implicated IL-23 in the induction of this response, and indeed, we show in this study that in the absence of IL-12p70, IL-23 is essential for the generation of IFN-γ-producing CD4⁺ T cells. That this compensatory response is unable to control experimental infection (7) indicates that the IL-23-induced IFN-γ response is not as potent as the IL-12p70-induced response, and our data demonstrate that the total number of Ag-specific CD4 T cells is significantly reduced when only IL-23 is available. A similar ability of IL-23 to compensate for the absence of IL-12p70 has been reported for the Toxoplasma model, in which exogenous IL-23 was able to improve control of pathogen burden in IL-12p40-deficient mice (33). That a compensatory IL-23 response may protect humans from Tb is suggested by the fact that although there are extensive reports of increased susceptibility to mycobacterial infection in humans lacking IL-12p40 or IL-12Rβ1 (34, 35, 36, 37, 38), there are no reports of susceptibility in humans lacking IL-12p35. The facts that mycobacterial stimulation of human macrophages results in IL-23 production (39) and that IL-23 has been shown to augment IFN-γ production in human T cells (8, 40) also support the potential for IL-23 to mediate protection in humans.

The essential requirement for IL-23 in the induction of the Ag-specific, IL-17-producing CD4⁺ T cell response to Mtb allowed us to determine the effect of IL-17 on the mycobacterial granuloma. We show in this study that in the absence of IL-17, the development of the mycobacterial mononuclear granuloma is not substantially altered. This failure is particularly surprising in view of the reported role of IL-17 as a mediator of inflammation in asthma (41), pneumonia (12), airway hypersensitivity (16), and neutrophil recruitment to the lung (30, 31, 32). It is also surprising because inflammatory responses are compromised in the absence of IL-12p40 (7), IL-23 (14), and IL-17 (16), but not in the absence of IFN-γ (42) or IL-12p35 (7, 14). It would therefore appear that although IL-17 is induced in response to Mtb infection, it plays little role in initial granuloma formation and long-term maintenance of the mononuclear nature of the granuloma. It may be the case that prolonged granuloma integrity requires IL-17, and we are investigating this possibility.

Although the ability of IL-17-producing CD4⁺ T cells to modulate the primary control of Mtb infection in the lung is limited, a high number of such cells is induced during Mtb infection. The majority of the literature supports the ability of IL-23 to augment the IL-17 production of already activated cells, with only a small proportion of naive cells progressing to an IL-17-producing phenotype in the presence of IL-23 (11, 15). In contrast, we show in this study that substantial numbers of naive TCR-transgenic T cells activated in the presence of cognate Ag and Mtb-activated DC, develop an IL-17-producing phenotype and that this is dependent upon IL-23. These data suggest that Mtb-activated DC are potent inducers of the IL-17-producing phenotype in naive CD4 T cells. It is intriguing to speculate that this ability to induce IL-17-producing cells combined with the dependence of experimental arthritis on IL-23 (19) is the basis for the association between mycobacterial exposure and arthritis (43, 44, 45).

The observed increase in the number of Ag-specific, IL-17-producing CD4⁺ T cells both in vivo (Fig. 7) and in vitro (Fig. 8) in the absence of IL-12p35 corresponds with the reported ability of IL-12p70 to limit IL-17 production in vitro (11) and the fact that in the absence of IL-12p70 (19) or IL-12Rβ2 signaling (46) there is an increase in IL-17-producing T cells. Our data clearly show that the absence of IL-12p70 during the initial activation of naive cells allows for a significant increase in the number of IL-17-producing CD4⁺ T cells, suggesting that the relative availability of IL-12p70 and IL-23 during the initiation of T cell activation will determine the eventual effector phenotype of the T cell. This ability of IL-12p70 to limit the activity of IL-23 may serve to mask the protective role of IL-23 in Mtb-infected C57BL/6 mice.

In conclusion, we show in this study for the first time that in the absence of IL-12p35, IL-23 is able to generate a moderately protective response to Mtb, and that this response is associated with the induction of IFN-γ-producing CD4⁺ T cells and the expression of IFN-γ-dependent, antimycobacterial genes. In contrast, although IL-23 is required for the induction of Ag-specific, IL-17-producing CD4⁺ T cells, the absence of these cells fails to modulate disease progression during primary Mtb infection.

The authors have no financial conflict of interest.